Re: [tcf-dev] Disassembly Service logic with Symbols

[Eugene Tarassov <eugenet@xxxxxxxxxx>]

Hi David,

Yes.
To be sure we don't start in the middle of an instruction, disassembler starts at a function entry or other suitable symbol, and it can return more than requested.

Regards,
Eugene

-----Original Message-----
From: tcf-dev-bounces@xxxxxxxxxxx <tcf-dev-bounces@xxxxxxxxxxx> On Behalf Of Wilson, David
Sent: Tuesday, September 1, 2020 9:10 AM
To: TCF Development <tcf-dev@xxxxxxxxxxx>
Subject: Re: [tcf-dev] Disassembly Service logic with Symbols

CAUTION: This message has originated from an External Source. Please use proper judgment and caution when opening attachments, clicking links, or responding to this email.


Hi Eugene - thanks so much for your quick answer.

So - just for further clarification - is it expected, that when symbols are loaded - and we ask for only 1 byte, i.e.:

Command: Disassembly disassemble "ContextID" Addr 1 {}

That we get MORE than one instruction back? We see this behavior when symbols are loaded, and I want to double check If this is expected or a bug. If it is expected - is the reasoning the same?

Best,

David



From: tcf-dev-bounces@xxxxxxxxxxx <tcf-dev-bounces@xxxxxxxxxxx>
Date: Monday, 31. August 2020 at 21:14
To: TCF Development <tcf-dev@xxxxxxxxxxx>
Subject: Re: [tcf-dev] Disassembly Service logic with Symbols
Hi David,

With variable instruction length (e.g. Intel ISA), disassembler must start at a function symbol and stop at the function end.
To avoid garbage output, it will stop at a function boundary regardless how many instructions are requested.
Clients are supposed to request more if needed and merge the data.

Regards,
Eugene

From: tcf-dev-bounces@xxxxxxxxxxx <tcf-dev-bounces@xxxxxxxxxxx> On Behalf Of Wilson, David
Sent: Monday, August 31, 2020 1:04 AM
To: 'TCF Development' <tcf-dev@xxxxxxxxxxx>
Subject: [tcf-dev] Disassembly Service logic with Symbols

CAUTION: This message has originated from an External Source. Please use proper judgment and caution when opening attachments, clicking links, or responding to this email.

Hi All,

I am wondering if I could get some clarification on how the disassembly command (in the disassembly service, disassembly.c) is meant to deal with cases when symbols are loaded.

The behavior I have noticed:

Command: Disassembly disassemble "ContextID" Addr 1 {}

This will return us at least 1 instruction (even though we've only specified one byte).

Then, we continue to ask for more instructions e.g.:

Command: Disassembly disassemble "ContextID" Addr 500 {}

And this will return us instructions up to the given size.


However - we have noticed, with symbols loaded, that this will reach a "limit" on the size of instructions we can ask for (i.e. we reach a point where we cannot ask for more.

This means even if we ask for: Command: Disassembly disassemble "ContextID" Addr 1000 {}

We still get the same number of instructions returned.


This happens because of the following block:

if (sym_addr_ok && sym_size_ok &&
                    sym_addr <= buf_addr && sym_addr + sym_size > buf_addr &&
                    sym_addr + sym_size <= buf_addr + buf_size) {
                buf_size = sym_addr + sym_size - buf_addr;
                mem_size = (size_t)buf_size;
            }

I am wondering if this behavior is expected, and if so - what the purpose of it is.


Kind Regards


David Wilson
Intel Deutschland GmbH
Registered Address: Am Campeon 10-12, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, http://www.intel.de
Managing Directors: Christin Eisenschmid, Gary Kershaw
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10-12, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de
Managing Directors: Christin Eisenschmid, Gary Kershaw
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928

_______________________________________________
tcf-dev mailing list
tcf-dev@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/tcf-dev